Jacobian warning for imputation of missing data?

Modeling
, ,
1

The line

for (i in 1:N) predictors_all[i,:] ~ multi_normal_cholesky(mu_predictors,
      diag_pre_multiply(sigma_predictors,Lrho_predictors));

is sometimes giving me the Jacobian warning: Left-hand side of sampling statement (~) may contain a non-linear transform of a parameter or local variable. If it does, you need to include a target += statement with the log absolute determinant of the Jacobian of the transform. Other times running code with the exact same logic I do not get this warning.

The matrix predictors_all is just a collection of predictor data with one observation per row and one predictor variable per column, formed from concatenating the vectors of observations for each of the individual variables. Missing observations in each vector were all replaced with parameters (prior to being concatenated into this matrix) to allow for imputation as part of the model.

Do I need to do anything in response to this warning, or is it not applicable in this context?

2

That’s indeed a false-alarm and you don’t need a jacobian.

1 Like
3

Thank you!